re: Sorting

Hello Everyone,

Apologies for not replying in a proper threaded fashion; I've followed 
this thread on the web archive of the mailing list, and so do not have a 
proper message to reply to; I've quoted Leigh Dodds' original message at 
the end.

I wanted to chime in to the discussion of sorting in SPARQL (the language) 
with another voice in favor of including it in the first version of the 
SPARQL spec. In particular, we believe there is significant value in 
having the expressiveness needed to write queries of the form: "give me 
the N latest objects of type Foo with property Bar". In the face of large 
data sets, such queries are only reasonable if the query language includes 
an ORDER BY facility that can work in conjunction with LIMIT. (For a very 
specific use case, consider the front page of http://del.icio.us which 
lists the 10 or so most recent bookmarks added.) 

Andy pointed me to the live version of the working draft where I see a 
first cut at adding ORDER BY (and OFFSET) to SPARQL; in the meantime, I 
just wanted to again echo my support for its inclusion. 

thanks kindly for considering it,
Lee


--- original message follows ---

From: Leigh Dodds <leigh@ldodds.com>
Date: Tue, 01 Mar 2005 13:58:07 +0000
Message-ID: <422474EF.3070504@ldodds.com>
To: public-rdf-dawg-comments@w3.org


Hi,

Whilst playing about with Sparql over the last month or two, I've been 
assuming that ordering of results is something that is "coming soon", 
i.e. in a later Working Draft.

However, now that I've actually looked, I see that Last Call is
approaching and that sorting/ordering of results is not in the
requirements list. In fact the only mention I can see of it
is this feedback from Andrew Newman [1] and a comment in the spec noting 
that the WG is aware of the relationship between sorting and limiting 
the number of returned results.

Is it the case then that Sparql will not include an ORDER BY or
similar clause? If not, then would it be possible to elaborate
why? (Or have I missed something and I'm being dull?)

I find this quite surprising if its not, as its a fundamentally
useful feature of any query language. Given the type support I'd
naively assumed that sorting would be a relatively straight-forward
extension.

Cheers,

L.

[1]. 
http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2004Oct/0010.html

Received on Thursday, 17 March 2005 05:25:13 UTC